Optimal Path-Planning via Waypoints for a Soaring UAV

In this paper, we solve the problem of optimal path-planning for an autonomous aerial glider that maximizes the cumulative reward collected by visiting predefined waypoints, while managing altitude loss. Previous UAV path-planning methods primarily utilize probabilistic models and learning-based path planning to handle uncertainties such as wind and thermal effects, which adds significant complexity due to the stochastic modeling. Our approach, on the other hand, utilizes the glider’s aerodynamic model, waypoint positions, available landing sites, and the associated rewards for each waypoint. We first show that the approximately optimal gliding strategy for coordinated turns and straight glide-path involves bang-bang style control inputs for bank angle and that it minimizes altitude loss. Leveraging this result, we precompute the altitude losses between waypoint pairs and to the nearest landing sites, which reduces the problem into an integer linear program inspired by the Traveling Salesman Problem (TSP). This path-planning strategy is then validated through simulation results, which demonstrate that our formulation efficiently produces globally optimal trajectories. This clearly illustrates the advantages of a dynamics-informed, deterministic optimization strategy for autonomous soaring gliders.